Surrogate model driving rapid transient analysis of vehicle induced girder-end displacement for suspension bridges

Abstract

The service life of restraint devices in suspension bridges is directly impacted by the girder-end cumulative longitudinal displacement, with vehicle loads being the primary factor contributing to excessive cumulative displacement. So, it is essential to analyze vehicle-induced girder-end displacements for optimized design, evaluation and maintenance of suspension bridge restraint devices. However, performing transient analysis on large-scale finite element models (FEM) to simulate the effects of vehicle loads presents significant challenges, including long computation times and high complexity. Therefore, a novel surrogate model for the rapid analysis of girder-end displacements induced by vehicles was proposed based on loads model and structural model dual dimension reduction. First, the structural input–output mapping relationship based on the structural dynamics was derived. Second, a dimension reduction loading strategy using equivalent vehicle loads was proposed. Furthermore, the reconstruction method of equivalent vehicle loads based on the input–output mapping relationship and Tikhonov regularization theory is determined. Finally, the structural model dimension reduction method was established based on the explicit iteration theory, and the computational process only needs to consider a few degrees of freedom associated with the girder-end longitudinal restraints. The surrogate model with dual dimension reduction of vehicle loads and the structural model enables rapid and precise calculation of girder-end displacements induced by stochastic vehicles. The real bridge analysis shows that the proposed method exhibits excellent computational accuracy and efficiency, with a computation time only 0.39% of that of the FEM transient analysis method. This represents a substantial reduction in computational workload for analyzing vehicle-induced girder-end displacements.